Interaction of Kinase-Interaction-Motif Protein Tyrosine Phosphatases with the Mitogen-Activated Protein Kinase ERK2

The mitogen-activation protein kinase ERK2 is tightly regulated by multiple phosphatases, including those of the kinase interaction motif (KIM) PTP family (STEP, PTPSL and HePTP). Here, we use small angle X-ray scattering (SAXS) and isothermal titration calorimetry (ITC) to show that the ERK2:STEP complex is compact and that residues outside the canonical KIM motif of STEP contribute to ERK2 binding. Furthermore, we analyzed the interaction of PTPSL with ERK2 showing that residues outside of the canonical KIM motif also contribute to ERK2 binding. The integration of this work with previous studies provides a quantitative and structural map of how the members of a single family of regulators, the KIM-PTPs, differentially interact with their corresponding MAPKs, ERK2 and p38α.


Introduction
The mitogen-activation protein kinases (MAPKs; ERK, p38 and JNK) are cytosolic serine/threonine-specific kinases that coordinate the cellular response to a range of extracellular stimuli. Their activity is tightly regulated by the coordinated action of activating kinases (the MAP kinase kinases), deactivating phosphatases (i.e., the kinase interaction motif protein tyrosine phosphatases or KIM-PTPs, the dual specificity MAPK phosphatases or MKPs, among others) and scaffolding/targeting proteins, which determine their subcellular location [1]. Once activated, the MAPKs phosphorylate a variety of substrates, which ultimately determine the cellular response [2]. Consequently, disruptions in MAPK signaling are correlated with multiple diseases, including cancer, autoimmunity and Alzheimer's disease, among others [3].
MAPKs have a bi-lobed architecture, with a five-stranded bsheet in the N-terminal lobe and six a-helices in the C-terminal lobe [4,5]. During the last two decades, cellular and structural studies have revealed that MAPKs contain a conserved docking groove, characterized by two distinct subsites (one charged, Y chg , and the second hydrophobic, W hyd ) [6][7][8]. Most regulators and substrates bind their cognate MAPKs at this groove via linear peptides, known as KIM-or D-site motifs (Kinase Interaction Motif), with the consensus sequence (R/K) 2-3 -X 2-6 -W A -X-W B (W is any hydrophobic residue) [9][10][11]. One family of regulators that bind MAPKs via a KIM is the KIM-PTPs (Kinase Interaction Motif -Protein Tyrosine Phosphatases) [11][12][13]. Members of this family include STEP (PTPN5, striatal enriched protein tyrosine phosphatase), PTPSL/PTPBR7 (PTPRR, PTP STEP-like) and HePTP (PTPN7, hematopoietic protein tyrosine phosphatase). They regulate ERK2 (extracellular signal-regulated kinase 2) and p38a by selectively dephosphorylating their activation loop tyrosines (active state complexes) and also by sequestering the kinases in the cytosol (resting state complexes).
Over the last years different groups, including ours, have studied the interaction of MAPK regulatory proteins, as well as substrates, with different MAPKs, especially ERK2, p38 and JNK [12][13][14][15][16][17]. However, while a plethora of cell biological data is available, much less systematic interaction data using quantitative methods, such as isothermal titration calorimetry (ITC), has been reported. This is of special interest, as recent studies have shown that while KIMpeptides derived from KIM-containing regulatory proteins and substrates readily discriminate between ERK2/p38 and JNK, they discriminate poorly between ERK2 and p38 [7]. We recently reported detailed ITC, NMR as well as small angle X-ray scattering (SAXS) data of the interactions between the MAPK p38a and the KIM-PTP family members [14,15]. Furthermore, we also compared the interaction the immune-specific KIM-PTP, HePTP, with p38a and ERK2 [16,17]. These studies showed that HePTP (residues 15-339) as well as the HePTP KIM-peptide (residues 15-31) consistently bind more tightly to ERK2 (,7-times and ,4-times, respectively) than p38a. To better understand the molecular determinants that allow regulatory proteins to discrim-inate between ERK2 and p38a, we investigated the interaction of PTPSL and STEP with ERK2 and then compared them with our previous study examining the same interactions with p38a. Our data shows that both PTPSL and STEP also bind more strongly to ERK2. Furthermore, we find that within the family both STEP and PTPSL interact with ERK2 and p38a more strongly than does HePTP. To determine if this discrimination between p38 and ERK2 is specific for this regulatory protein family, we also compared the interaction of a MAP Kinase Phosphatase (MKP; DUSP16) MAPK binding domain with p38a [18] and ERK2. In contrast to what we observed for the KIM-PTPs, DUSP16 binds both MAPKs with similar affinities, demonstrating that other families of MAPK regulator proteins likely have distinct discrimination factors between the MAPKs. Finally, we correlated these differences with the low-resolution structures of the complexes using small angle x-ray scattering (SAXS) experiments. Collectively, our data is providing a coherent understanding of how subtle differences between the KIM-PTPs contribute to MAPK selectivity and, in turn, MAPK signal fidelity.

Isothermal Titration Calorimetry
ITC experiments were performed at 25 uC using a VP-ITC microcalorimeter (Microcal Inc.). Titrant (10 mL per injection) was injected into the sample cell over a period of 20 seconds with a 250 second interval between titrations to allow for complete equilibration and baseline recovery. 28 injections were delivered during each experiment and the solution in the sample cell was stirred at 307 rpm to ensure rapid mixing. To determine the thermodynamic parameters (DH, DS, DG) and binding constants (K) of the ERK2:PTPSL KIM , ERK2:PTPSL, ERK2:STEP and ERK2:PSChimera interactions, PTPSL KIM , PTPSL, STEP and PSChimera were titrated into ERK2. To determine the same parameters for the ERK2:STEP KIM , ERK2:PTPSL KIMKIS and ERK2:STEP KIMKIS interaction, ERK2 was titrated into the peptides. Titrants were chosen based on the amount of purified protein/peptide available for the experiment. For all titrations the concentrations were slightly varied to ensure independence of the K d from the protein/peptide concentrations. Data were analyzed with a one-site binding model assuming a binding stoichiometry of 1:1 using Origin 7.0 software. A nonlinear least-squares algorithm and the titrant and sample cell concentrations were used to fit the heat flow per injection to an equilibrium binding equation, providing values of the stoichiometry (n), change in enthalpy (DH) and binding constant (K). All data were repeated in triplicate.

Small angle X-ray scattering
To form the ERK2:STEP resting-state complex, purified, unphosphorylated ERK2 was incubated in a 1:1 molar ratio with purified STEP 213-539 at room temperature for 30 minutes and purified using SEC (Superdex 75 26/60 [GE Healthcare]; SAXS buffer). Data for the ERK2:STEP complex were recorded on samples at 0.7, 1.2 and 1.7 mg/ml. All samples were prepared within 48 hours of data acquisition and stored on ice at 4 uC. All samples were filtered (0.02 mM filter, Whatman) immediately prior to data collection. All data was recorded at Beamline X9A at the National Synchrotron Light Source (NSLS) using a Dectris Pilatus 300k (3.4 m distance from the sample for SAXS) and a Photonic Science CCD (0.47 m from the sample for WAXS) detector. 20 ml of sample was continuously flowed through a 1 mm diameter capillary and exposed to an X-ray beam for 30 seconds. Normalization for beam intensity, buffer subtraction and merging of the data from both detectors were carried out using PRIMUS [20]. A Guinier approximation, I(q) = I(0)exp(-q 2 R g 2 /3), where a plot of ln[I(q)] and q 2 is linear for q,1.3/R g (q is the scattering vector q = 4psin(h)/l; l, wavelength of X-rays and 2h is the angle between the incident beam and the scattering X-rays) was performed on at least four independent scattering trials and averaged to determine the radius of gyration (R g ). The linearity of the Guinier region and the intensity at zero scattering angle, I(0), were used to validate that all samples were monodisperse in solution. I(0)/c, where c is concentration, was consistent for all measurements. GNOM [21] was used to determine the pairdistribution function, P(r), for the ERK2:STEP resting-state complex using data collected on the 1.7 mg/ml sample. Twentyfour envelopes were generated using GASBOR [22] and were aligned and averaged using the DAMAVER program suite [23].

KIM-PTPs bind ERK2 more tightly than p38a
We used ITC measurements to determine the binding affinity of STEP and PTPSL for unphosphorylated ERK2 (Table 1; Figure S1). ITC showed that STEP binds ERK2 ,12-times more tightly than p38a. Indeed, the ERK2:STEP complex is the tightest binding complex of all MAPK:KIM-PTP complexes tested. This complex has a ,3-times lower K d (tighter binding) than the next tightest complex, ERK2:PTPSL, and a nearly 70fold greater affinity than the weakest complex, p38a:HePTP [15]. Like STEP, PTPSL binds ERK2 more tightly (,3.5-fold) than p38a. We also tested peptides that include the kinase specificity sequence (KIS) regions, as we [14,15] and others [12] have shown that KIS residues can interact directly with ERK2 and p38a. Each STEP (STEP KIM , STEP KIMKIS , STEP; Figure 1A) and PTPSL (PTPSL KIM , PTPSL KIMKIS , PTPSL; Figure 1A) domain has a higher binding affinity for ERK2 than for p38a [14]. This increased affinity for ERK2 was also observed for each domain of HePTP [15,17]. Interestingly, the tightest p38a:KIM-PTP complex, p38a:PTPSL (K d ,380 nM), is equivalent to the weakest ERK2:KIM-PTP complex, ERK2:HePTP (K d ,370 nM) [15,17]. Therefore, when compared to p38a, ERK2 has a higher affinity for each KIM-PTP and KIM-PTP domain (KIM, KIMKIS, KIM-PTP), showing that the increased interaction strength is clearly driven by the MAPK rather than the regulatory protein.
We also generated a chimeric protein in which we fused the PTPSL KIMKIS to STEP CAT (PS chimera ; Figure 1A) [14]. This was done in order to test the contribution of the STEP CAT to the MAPK:protein interaction, as we previously showed that the STEP CAT plays a role in binding p38a [14]. The PS chimera binds marginally more tightly to p38a than any of the WT-KIM-PTPs, but binds to ERK2 with about equal binding strength as STEP ( Table 1) [14].
The ERK2:STEP resting-state complex is compact ITC measurements showed that KIM-PTPs interact more strongly with ERK2 than with p38a. In order to identify if these increased interaction strengths are correlated with structural changes we used SAXS to determine the overall shape of these complexes in solution. We recently showed that the ERK2:HePTP resting-state complex is highly extended in solution with a radius of gyration (R g ) of 33.3 Å and a maximum dimension (D max ) of 110 Å [16]. Here, we used SAXS to determine the size and shape of the ERK2:STEP resting-state complex ( Figure 1B). The ERK2:STEP complex was purified by SEC and SAXS data were recorded at Beamline X9A at the National Synchrotron Light Source (Brookhaven National Laboratories). Guinier analysis of six independent SAXS samples was used to calculate an R g of 30.660.4 Å , showing that the R g of the ERK2:STEP resting-state complex is ,3.0 Å smaller than that of the ERK2:HePTP restingstate complex (Table 2; Figure 1C; HePTP 324 residues; STEP 326 residues) [16]. To analyze this difference in more detail, we determined the ab initio molecular envelope of the ERK2:STEP resting-state complex ( Figure 1C). A D max of 95 Å was determined by analysis of the P(r) function, which is ,15 Å shorter than that of the ERK2:HePTP complex (Figure 2) [16]. Thus, like the p38a:STEP resting-state complex [14], the ERK2:-STEP resting-state complex is compact (Figure 2), and suggests that the STEP CAT interacts directly with ERK2. Consistent with this hypothesis, STEP binds ERK2 11-times more tightly than the STEP KIM peptide ( Table 1). Based on our ITC data, both the STEP KIS and the STEP CAT contribute to the binding between ERK2 and STEP, as the STEP KIMKIS binds to ERK2 ,4.5-times more tightly than the STEP KIM and ,2.5-times less tightly than full-length STEP ( Table 1).
The calculated R g and D max of the ERK2:STEP resting-state complex are highly similar to those determined for the p38a:STEP resting-state complex (Figure 2). Using NMR spectroscopy, we previously determined that STEP CAT interacts directly with p38a; specifically, helix a-29 of STEP binds between the p38a activation loop and the MAPK-insert [14]. Our SAXS and ITC data suggest that STEP CAT also interacts with ERK2. Although there are a few structural differences between ERK2 and p38a in this region (e.g., the activation loop is six residues longer in ERK2), nearly all residues that that we showed participate in the p38a:STEP CAT interaction are conserved between ERK2 and p38a [14]. Taken together, it is very likely STEP CAT interacts with ERK2 via a mechanism very similar to that observed for p38a.

The interaction of PTPSL with ERK2 and p38a is not conserved
While HePTP interacts in an extended fashion with ERK2 and p38a, our data shows that STEP associates with these MAPKs to form more compact, globular complexes. Recently, Balasu et al. used cross-linking, mass-spectrometry (MS) and molecular modeling to generate a structural model for the ERK2:PTPSL restingstate complex, in which they predict that the complex is compact, with ERK2 and PTPSL associating in a side-by-side manner [24]. In this model, PTPSL CAT binds ERK2 in an orientation in which the catalytic site is positioned near the ERK2 phosphorylation loop. This is different from the p38a:PTPSL complex, which we showed is extended in solution [14]. Despite exhaustive attempts to collect SAXS data on the ERK2:PTPSL complex, the samples persistently exhibited the typical hallmarks of aggregation and thus, could not be used for further analysis. However, using ITC, we show that full-length PTPSL binds ERK2 ,4-times more tightly than the PTPSL KIM peptide ( Table 1). This additional binding strength is not a result from the interaction of PTPSL KIS residues with ERK2, as the PTPSL KIMKIS binds less tightly than the PTPSL KIM . Therefore, it is the interactions between PTPSL CAT and ERK2 that lead to the increased binding strength observed for the ERK2:PTPSL complex. Thus, our ITC data support the model Balasu et al. proposed, in which PTPSL CAT directly interacts with ERK2 [24].
Unlike the KIM-PTPs, DUSP16 binds p38a and ERK2 with similar affinities The KIM domains of KIM-PTPs are intrinsically disordered in solution [14,15]. Thus, they only adopt a single conformation when they bind their respective MAPKs. This is different from the MAPK-binding domains (MKBD) of MAPK phosphatases (MKP; dual specificity phosphatases; DUSPs) [25,26]. In MKPs, the KIM sequence is part of a stable, folded domain and thus adopts a single, ordered conformation in both the unbound and MAPKbound states [27,28]. Using a combination of biophysical studies, we recently showed how the DUSP16 MAPK-binding domain interacts with p38a [18]. Our NMR data showed that the MAPKbinding domain of DUSP16 interacts at multiple positions with p38a. While some of them overlap with those used for KIM-PTPs (e.g. interaction with W B and Y chg ) others are distinct (helix a4) and based on our results important for the interaction of MAPKbinding domain of DUSPs with MAPKs. The ITC measurements reported a K d of ,900 nM. In order to determine if DUSP16, like the KIM-PTPs, discriminates between p38a and ERK2, we carried out similar ITC measurements between the DUSP16 MKBD and ERK2. ITC reported a K d of ,1300 nM ( Table 1; Figure S1), which is only ,1.4-fold weaker than the interaction with p38a. Thus, while the KIM-PTPs clearly exhibit an increased affinity for ERK2 over p38a, DUSP16 binds these MAPKs with similar affinities.

Discussion
Over the last years it has become apparent that the KIM-motifs from most regulatory proteins readily discriminate between JNK and p38/ERK, however they discriminate much more poorly between p38 and ERK. For example, in a recent study, JNK-  [14,15,17], but are shown here to provide a better comparison of the P(r) functions). doi:10.1371/journal.pone.0091934.g002 specific KIM peptides were only specific to JNK, with no detectable binding of these JNK-specific peptides by p38/ERK. In contrast, KIM peptides specific for either p38 or ERK2 exhibited discrimination factors of less than 4-fold between p38 and ERK2 [7]. Here, we show that while a subset of p38/ERK2 regulators, the KIM-PTPs, interact productively with both p38a and ERK2, their affinities for ERK2 are higher ( Figure 3). Specifically, the KIM peptides of the KIM-PTPs exhibit differences in affinity for ERK2 versus p38a between 4.0 (HePTP) and 6.5 (STEP). Furthermore, for the interaction between MAPKs and the KIM-PTP proteins (i.e., not just the KIM-peptides), the fold differences in affinity become much larger, to as high as 12 for STEP. This shows that residues outside the KIM contribute to the overall interaction strength of MAPKs for their regulatory proteins ( Figure 3). It also demonstrates that the increased interaction strength is clearly driven by the MAPK rather than the regulatory protein. The amino acid sequence of p38a is ,45% identical to ERK2 and multiple residues that line the KIM hydrophobic pocket, especially around W B (p38 WB ERK2: G110E/A111T/ N115K) and the CD site (p38 CD ERK2: D124N/Q128C/ E160T/D161T) are not conserved between the two MAPKs. Thus, these sequence differences are the likely reason for the reduced affinity of the regulatory proteins for p38a versus ERK2.
In contrast, and more consistent with the previous study, the interaction of the KIM in DUSP16 exhibits almost no difference in affinity for p38a and ERK2, with a difference in K d of only ,1.4-fold. Together, our data shows that while KIM peptides typically poorly discriminate between ERK2 and p38a (as compared to JNK), significant discrimination for these two MAPKs can be seen for at least a subset of regulatory proteins, the KIM-PTPs. Collectively, these studies are providing a more coherent understanding of how differences in the family of KIM-PTPs with ERK2 or p38a contribute to MAPK selectively and, in turn, MAPK signal fidelity.